Accepted Manuscript

Electrical, thermal and mechanical properties of poly-etherimide epoxy-diamine blend

N. Halawani, J.L. Augé, H. Morel, O. Gain, S. Pruvost

PII: S1359-8368(16)31597-9
DOI: 10.1016/j.compositesb.2016.11.022
Reference: JCOMB 4710

To appear in: Composites Part B

Received Date: 11 August 2016

Revised Date: 6 October 2016
Accepted Date: 11 November 2016

Please cite this article as: Halawani N, Augé JL, Morel H, Gain O, Pruvost S, Electrical, thermal and
mechanical properties of poly-etherimide epoxy-diamine blend, Composites Part B (2016), doi: 10.1016/
j.compositesb.2016.11.022.

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to
our customers we are providing this early version of the manuscript. The manuscript will undergo
copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please
note that during the production process errors may be discovered which could affect the content, and all
legal disclaimers that apply to the journal pertain.
 ACCEPTED MANUSCRIPT

Epoxy – polyetherimide, a homogenous phase separated blend,

studied to show the effect of phase separation phenomenon in an
organic-organic blend on the thermal, mechanical and electrical
properties. The presence of the thermoplastic nodules decreased
the permittivity values and the simulation prediction showed similar
results with a shift to higher values.

PT
RI
U SC
AN
M
D
TE
EP
C
AC
 ACCEPTED MANUSCRIPT
1

Electrical, thermal and mechanical properties of

Poly-etherimide epoxy-diamine Blend
N. Halawani a,b, J.L. Augé a, H. Morel a, O. Gain b and S. Pruvost *b

PT
a
Univ Lyon, INSA Lyon, UMR CNRS 5005, AMPERE, F- 69622, Villeurbanne, F rance
b
Univ Lyon, INSA Lyon, UMR CNRS 5223, IMP Ingénierie des Matériaux Polymères, F-69621, Villeurbanne, F rance

*corresponding author: sebastien.pruvost@insa-lyon.fr

RI
Abstract—This work deals with the study of thermoset-thermoplastic blend forming an epoxy-amine/poly-

etherimide phase separated to assess the dielectric and thermal performances. These materials would be good

SC
candidates to replace the passivation layer in semiconductors, particularly ones used as switches in power

electronic applications. Polyimide and Parylene are usually quoted and studied. Mixtures based on polymers

U
would be a novel candidate that can manage to be an insulator for the system. The modified and non-modified
AN
systems where characterized by transmission electron microscopy, scanning electron microscopy, differential

scanning calorimetry, thermogravimetric analysis, dynamic mechanical analysis, dielectric analysis and
M

analytical simulation was carried out on dielectric measurements. These complementary techniques were used

first to investigate the presence of the phase separation phenomenon and second to quantify the separated
D

nodules size. The effect of this phase separation was examined and showed enhancement in the dielectric
TE

values compared to the pure epoxy system. It was finally simulated to show a close assumption of what is

found experimentally.
EP

Keywords: A. Thermosetting resin; organic-organic blend; A. thermoplastic; B. electrical properties

1. INTRODUCTION
C

In power IGBT (insulated-gate bipolar transistor) modules, the organic insulating materials as polyimide and silicone
AC

are mainly present. Polyimide is deposited by spin coating while silicone gel is deposited by molding under vacuum.

These materials play a significant role in the improvement of the package voltage ratings under high tension and

even if the temperature range for using silicone is limited for around 200 °C under a continuous use. On the other

side, the softness and high electrical properties of silicones render it interesting. Furthermore, under high voltage

partial discharge can give rise to the breakdown of the device. In an alternator, the reduction of the partial discharges

effect was carried out by incorporating mica flakes in the insulating material to play the role of a barrier against
 ACCEPTED MANUSCRIPT
2

discharges [1]. The goal of this paper is to present a novel material that could be another good candidate used as

passivation layer or the encapsulation layer in the power modules.

Generally, composites are made up of materials combinations differing in composition, where the individual

PT
constituents retain their separate identities. Separate constituents acting together give the final morphology and

property of the composite.

RI
Polymers composites are one of the most famous composites, such as metal matrix composites (MMC) or

ceramic matrix composites (CMC), commercially used and studied in several domains such as electrical engineering

SC
due to their good dielectric properties, light weight and ease of production [2-8]. One of the studied kinds of

polymer composites is epoxy filled with inorganic micro or nano particles; it is an organic-inorganic blend. Epoxy

U
networks are extensively used as highly crosslinked materials in many applications due to their special performance,

such as good mechanical, thermal and electrical properties as well as low thermal expansion coefficient compared to
AN
silicon gels. They are now widely used in many applications such as adhesives, coatings, construction, and

composite matrices in the aerospace and electronics industries. Many investigations are done to show and compare
M

the effects of nano and micro fillers on the final electrical characteristics of the epoxy network [9-14]. With the

addition of micro-fillers to the epoxy network, the relative permittivity of the epoxy micro-composite is always
D

higher than that of the pure epoxy network [11,12]. Nanocomposites with relatively low filler concentrations, i.e.,
TE

less than 1vol.% , pointed out a relative permittivity ε'r which is lower than that of both the neat epoxy network and

the filler but with an increase in the loss values [14-16]. They also presented improvements in both thermal and
EP

electrical properties [14,17-19]. As the filler concentration increases, the real permittivity of nanocomposites can be

much higher than the pure epoxy network. Behavior of the nanocomposites depends on both the permittivity and the
C

concentration of the fillers.

AC

Another kind of polymer blend is the mixture of polymer with another organic substituent, organic-organic blend

wherein this mixture can produce miscible or immiscible systems. Immiscible block copolymers are widely studied

in a way many parameters can be adjusted and other organic structures can be added in order to change the

miscibility of the mixture [20-22]. Nonetheless, miscible mixtures of polymer and organic mixtures such as

thermoplastic-thermoset blends have gained importance in the previous decades. Adding a thermoplastic to the

thermoset network (with phase separation) can improve some of the mechanical properties of the thermoset, such as
 ACCEPTED MANUSCRIPT
3

network resistance toughness [23]. In most cases the thermoplastic-thermoset blends are prepared starting from a

homogenous solution of the thermoplastic polymer in the prepolymer and during the polymerization of the thermoset

a phase separation occurs in the blend at a given epoxy conversion rate. This mechanism is known as reaction-

induced phase separation [24,25] and may lead to several morphological types such as particulate, bicontinuous, or

PT
phase inverted, an illustration schema is presented in Fig. 1 The separation proceeds via the spinodal demixing (SD)

mechanism and thus the obtained system is homogenous. Many attempts have been made to modify epoxy networks

RI
with high performance engineering thermoplastics that are characterized by high Tg and high toughness, such as

polysulphone (PSF) [26,27], Poly-ethersulfone (PES) [28-30] and poly-etherimides (PEI) [23,31-36]. Commercial

SC
products exhibit large variation in average molar masses of engineering thermoplastics and thus provide different

modification to the epoxy properties [24].

U
Investigation of phase separation behavior in PEI-modified epoxy networks was done extensively since 1989 [31].
AN
Tetrafunctional epoxy resins tetraglycidyl 4,4′-diaminodiphenylmethane (DGDDM) was blended with PEI

toughening agents and using 4,4′-diaminodiphenylsulfone (DDS) as curing agent [37]. SEM micrographs reveal
M

spherical PEI domains in epoxy network at 5 wt.% PEI used and these domains became larger as the PEI contents

were increased up to 20 wt.%. Bonnet et al. worked on Bisphenol A diglycidyl ether (DGEBA) prepolymer with PEI
D

and confirmed these observations associated to an improvement in the fracture toughness of the blend at room
TE

temperature due to the presence of PEI [23,36]. The improvement increases along with the increase of the PEI

percentage blended with the epoxy network [38].

EP

This study deals with the effect of phase separation phenomenon in an organic-organic blend on the thermal,

mechanical and electrical properties. The chosen thermoset matrix is an epoxy-amine network in which PEI is added
C

to obtain the phase separation with nodules of PEI inside the thermoset matrix. The chosen epoxy system is a model
AC

system with controlled reactivity [39]. To insure the stability of the epoxy system along a wide range of temperature,

the epoxy system chosen have high glass transition Temperature (Tg) of around 160 °C, and the thermoplastic, PEI,

have higher Tg around 220 °C.

 ACCEPTED MANUSCRIPT
4

2. EXPERIMENTAL

2.1. Materials Evaluated

The epoxy prepolymer used was Diglycidyl ether of bisphenol A, liquid at room temperature, .

(Araldite LY556 from Huntsman). The curing agent was an aromatic diamine, 4,4’-methylenebis-[2,6-

PT
diethylaniline], M-DEA supplied by Lonzacure™. The diamine was added in the stoichiometric ratio, epoxy to

amine hydrogen groups equal to 1, in pure and modified mixtures. The thermoplastic used was Polyetherimide, PEI

RI
Ultem 1010, supplied by General electric. The chemical structures are listed below in Table 1.

SC
To prepare the neat network (without thermoplastic), the diamine was added to the epoxy prepolymer heated at 90

°C and stirred under vacuum. The PEI modified mixture was prepared in a three-step process: first PEI was heated at

150 °C for several hours to remove possible humidity in the powder, then dissolved in the epoxy prepolymer at 140

U
°C, and finally the diamine was added at 90 °C. The two mixtures were stirred under vacuum and poured in silicon
AN
preheated molds. For insuring a complete crosslinking process, the curing process was 4 h at 135° C followed by 2 h

at 200 °C of post-curing for pure epoxy network and 4 h at 160 °C followed by 2 h at 185 °C of post-curing for PEI
M

modified mixture. Then the mixtures were left to cool down to room temperature. Precuring durations where chosen

larger than the vitrification time to be sure that most of the microstructures are developed isothermally. The
D

precuring temperature determines the final morphology of the PEI thermoplastic into the epoxy network [23,40]. PEI
TE

pure films were prepared by pressing PEI powder at 270 °C under 30 bars. The obtained film has a thickness of 0.2

mm.
EP

For the designation of the systems, the term DM corresponds to the epoxy-amine (DGEBA-MDEA) system

without thermoplastic, which is also called neat network, and DM10PEI denotes the system DGEBA-MDEA

modified with 10 wt.% of PEI.

C
AC

2.2. Techniques

Phase separation as well as size distribution of the PEI nodules in the epoxy network was observed by Scanning

Electron Microscopy (SEM) Microscope ESEM-FEG FEI XL30 at CLYM (Centre LYonnais des Microscopies).

Fractured Samples were etched with a 1 wt.% solution of potassium permanganate in a 5:2:2 volume mixture of

concentrated sulfuric acid-phosphoric acid-distilled water. After etching, the samples were sequentially washed in

aqueous sulfuric acid, hydrogen peroxide (100 volumes), water, and finally acetone [41]. All samples were then
 ACCEPTED MANUSCRIPT
5

sputter coated with gold before SEM examination. Size distribution of the dispersed nodules was obtained using

ImageJ software.

TEM analysis was performed on Philips CM120 transmission electron microscope (at the Centre Technologique

PT
des Microstructures CTµ of the University of Lyon) with an accelerating voltage of 80 kV. Epoxy samples were

trimmed using an ultra-microtome and the slices (60–70 nm in thickness) were placed onto 300 mesh copper grids

RI
for observation.

SC
Glass transition temperature, determined as midpoint temperature Tg, midpoint, of the epoxy-amine systems was

measured using Differential Scanning Calorimetry (DSC) TA Q20. Experiments were carried out on a 5 mg sample

U
under controlled atmosphere of helium in a temperature range from -100 to 300 °C at a heating rate of 10 °C/min.

Temperature and heat flow were calibrated using indium.

AN
The thermogravimetric analysis (TGA) were performed using TA Q500 analyzer from 25 °C to 800 °C at a heating
M

rate of 10 °C/min in both air and nitrogen atmospheres.

D

Dynamic mechanical analysis (DMA) was carried out using a TA Q800 instrument with a heating rate of 3 °C/min.
TE

Tensile testing clamp for rectangular samples was used. Samples tested are of a rectangular shape with a thickness

between 0.5 and 1 mm, a length of 20 mm and a width of 10 mm. The tensile strain is applied at a frequency of 1 Hz
EP

and its value is equal to 0.01 %. Storage modulus E’, Loss Modulus E’’ and the loss factor tan δ were measured

during temperature ramps from -100 °C up to 300 °C with a rate of 3 °C/min.

C
AC

Novocontrol Alpha-A analyzer (BDS Novocontrol) equipped with the Novocontrol quatro temperature controller

(BDS 1330) for temperature control is used to characterize the dielectric properties of samples under a temperature

range from -100 to 250 °C with a heating rate of 2 °C/min and Vrms = 3V. Samples used are of circular shape with a 4

cm diameter and a thickness between 0.5 and 1 mm. Two gold electrodes, of 30 mm in diameter, were sputtered on

both sides using BAL-TEC SCD 005 Cool Sputter Coater. The model used for measurement is metal-insulator-metal

(MIM) capacitor. The complex dielectric constant of samples was measured over a frequency range from 107 to 10-1
 ACCEPTED MANUSCRIPT
6

Hz.

3. RESULTS AND DISCUSSION

3.1. Morphology of cured epoxy-PEI blend

PT
SEM images were performed on the neat network and 10 wt.% PEI in the epoxy-amine network and shown in Fig.

2(a) and (b) respectively. The aim was to observe if the phase separation phenomenon occurs between PEI and the

RI
epoxy network. The dark part corresponds to the epoxy rich phase and the bright parts correspond to the PEI rich

phase. It can be seen that the phase separation produced a well dispersed PEI phase in a continuous epoxy network.

SC
Due to the low reactivity of the DGEBA-MDEA system, the separation proceeded via the spinodal demixing (SD)

mechanism [32].

U
The size distribution of the nodules inside the epoxy-amine network is presented in Fig. 2(b) The separated
AN
nodules have a spherical or ellipsoidal form with an average diameter between 2 and 3 µm. From literature

[31,33,34], a similar morphology was observed for the same range of weight percentage of PEI but within a

tetrafunctional amine hardener. Temperature of precuring affects the nature of the reactions in an epoxy-
M

amine/thermoplastic mixture and governs the structure/morphology of the resultant blends. It has been shown that
D

slower epoxy curing rate and specifically precuring ones, where long gel times were obtained, resulted in non-

spherical thermoplastic domains [23].

TE

Fig. 3 shows a TEM image of the DM10PEI sample in which the phase separation phenomenon observed by
EP

means of SEM. The dark part corresponds to the PEI phase and the bright part corresponds to the epoxy phase. An

interphase is seen surrounding the PEI separated nodules having a thickness of 20-50 nm. This interphase is not
C

found to be present surrounding all the nodules.

AC

3.2. Thermal Properties

Thermogravimetric diagrams under air and nitrogen atmosphere of the three systems are displayed in Fig. 4 (a)

and (b) respectively. The system is considered stable until 5 % of its mass is lost. The neat epoxy system showed

good thermal stability up to 300 °C under air atmosphere and up to 350 °C under nitrogen atmosphere with a Tmax

peak of the derivative weight change at 380 °C and 400 °C respectively. Afterwards, chain scission of the network
 ACCEPTED MANUSCRIPT
7

takes place yielding combustible gases, water, amines, and gaseous aromatic compounds, etc. [42]. The lower

temperature degradation observed under air can be explained by the presence of the oxidation phenomenon. The pure

PEI system showed similar stability under both atmospheres up to 525 °C with a difference in the residual mass. The

addition of PEI in the epoxy network did not significantly affect its thermal stability upon modifying it with 10 wt.%

PT
of PEI.

3.3. Glass Transition Temperature

RI
Fig. 5 (a) shows DSC measurements realized on the three studied systems (Epoxy network, Epoxy-PEI network

and PEI). PEI film showed a transition corresponding to the glass transition at 216 °C. DM sample showed also a

SC
single transition corresponding to the glass transition temperature of epoxy-amine network at 164 °C. The DM10PEI

sample exhibited a transition similar to the latter with the same value of specific heat constant having values of 0.15

U
J/ (g. °C). Both DM and DM10PEI did not present an exothermic peak on the first run, which insures full cross-
AN
linking of the epoxy amine network. The same temperature of the glass transition obtained on the second run

(increasing temperature) confirmed this result. The glass transition corresponding to the thermoplastic phase in the
M

epoxy network is not marked due to its low concentration in the system. A derivative of the heat flow of the

DM10PEI system with respect to temperature is shown in Fig. 5(b) where the second transition of the system can be
D

observed at 216 °C corresponding to the glass transition temperature of the PEI. Phase separation phenomenon is
TE

confirmed by the presence of two glass transitions in the DM10PEI system corresponding to the two separated

systems Epoxy-MDEA and PEI. Both the transition of the epoxy and the PEI systems were not shifted upon the

presence in the DM10PEI network which proved a total phase separation of the system.
EP

Dynamic mechanical data are plotted in Fig. 6 for DM, DM10PEI and PEI. The storage moduli curves are
C

depicted in Fig. 6(a); DM showed a constant stable value over a wide range of temperature and up to α relaxation
AC

temperature (Tα epoxy =183 °C) of the epoxy network where the decay of the storage modulus was observed. The

storage modulus value is higher for PEI than for DM. At 226 °C, the relaxation α associated to the glass transition of

PEI Tα (PEI) appeared. The 10 wt.% PEI/epoxy blend followed the same evolution as that of the neat system with

the same decay phenomenon corresponding to Tα (epoxy).

Tan delta curves in Fig. 6(b) underline that the neat epoxy network gives two loss peaks. One broad peak at low
 ACCEPTED MANUSCRIPT
8

temperature with a maximum at -60 °C was assigned to the β relaxation (Tβ) attributed to the movement of the

hydroxyether units(-O-CH2-CH(OH)-CH2-) created by the crosslinking reaction and the initial epoxy prepolymer

[43]. A second sharp peak corresponding to Tα associated to the glass transition temperature of epoxy network is

observed at 183 °C. The α relaxation in epoxy networks depends on the nature of the epoxy prepolymer as well as

PT
the hardener [12,29,41-44]. One sharp peak with a maximum at 229 °C is observed and ascribed to Tα associated to

the glass transition of PEI.

RI
The phase separation phenomenon, suggested by SEM images, for the 10 wt.% PEI/epoxy blend (so-called

SC
DM10PEI) is confirmed in the DMA measurements by the presence of two high-temperature relaxations associated

to the epoxy network (Tα1) and the PEI phase (Tα2) respectively. DM and DM10PEI exhibited the same temperature

U
for α relaxation of the epoxy network proving that the epoxy network is totally separated from the thermoplastic.

Alternatively, Tα2 is slightly lower than pure PEI by 6 °C. This shift could be attributed to two different phenomena.
AN
First, a small amount of epoxy-amine is miscible in the PEI phase plasticizing it. Secondly, the presence of free

volume, in which it has been shown that blending of polymers generate additional free volume, which can lower Tg
M

by reducing the energy necessary for segmental motion [23,45]. In the next part, these two possibilities will be

discussed.
D
TE

In addition to SEM images, more information about the morphology of the system can be obtained by studying the

heights of loss peaks and, more specifically, the ratio of these heights (h2/h1). This value appears to be very sensitive
EP

not only to the volume fraction but also to the shape of final morphology through the choice of the curing

temperature. The characteristics of the two high temperature loss peaks, their maximum temperature (Tα1, Tα2), their
C

amplitudes (h1, h2) , the ratio of these heights (h2/h1) as well as the values of the storage modulus before and after the
AC

α relaxation temperature are collected in Table 2. The amplitude of the peak corresponding to the epoxy network (h1)

remains constant with the addition of PEI. On the other hand, the amplitude of the peak corresponding to the PEI into

the epoxy network is significantly reduced due to the low percentage of PEI present in the epoxy network [38].

The ratio (h2/ h1) can be used to predict qualitatively the kind of morphology generated in thermoplastic/thermoset

blends and to know if the structure of the network is inverted or not. In this studied case of DM10PEI, its ratio is
 ACCEPTED MANUSCRIPT
9

equal to 0.14. This value corresponds to the morphology of the separated nodules in the epoxy network as described

by Girard-Reydet et al. [23] and confirms the particulate morphology observed by SEM. The particulate

morphology of PEI into the epoxy network did not affect the value of storage modulus of the epoxy network.

Nevertheless it have been noted that the PEI nodules formed an obstacle to the propagation of mechanical crack and

PT
may participate to the crack blunting effect [23].

RI
3.4. Dielectric Properties

Dielectric measurements were carried out on the three systems to monitor the influence of the addition of the

SC
thermoplastic material into the epoxy network; their dielectric response is presented in Fig. 7(a). The real (ε'r) and the

imaginary (ε''r) parts of the dielectric relative permittivity were derived from measurements under a 3 Vrms sinusoidal

U
applied voltage.
AN
Regarding the dielectric response at a constant frequency of 1 kHz versus temperature Fig. 7(a), the neat epoxy

network showed relative permittivity ε'r values starting from 4.5 at low temperatures and increasing up to 6.5 at
M

temperatures higher than ambient temperature. A small increase with a maximum value around 183 °C corresponds

to the α relaxation of the epoxy system. PEI has relative permittivity ε'r values lower by 30% than that of the neat
D

epoxy network and constant on the whole range of temperatures. Choudhury et al. observed similar results for the
TE

relative permittivity ε'r of PEI [46]. The introduction of 10 wt.% of PEI into the epoxy network led to a decrease of

around 15 % of the relative real permittivity ε'r. The shape of the curves was not influenced by the addition of PEI

but it was only shifted to lower values.

EP

Imaginary relative permittivity values ε''r Fig. 7(b) of epoxy network two main transitions are identified. The first
C

transition at -30 °C corresponds to β transition associated to the movement of the hydroxyether units created by the
AC

crosslinking reaction and the initial epoxy prepolymer. The second main transition starting at 170 °C corresponds to

α transition of epoxy network that appears in the form of a shoulder superimposed by the conduction phenomenon;

this transition was also observed in DMA measurements. PEI network exhibited one main α relaxation at 240 °C

associated to the glass transition of PEI. The introduction of 10 %wt of PEI into the epoxy network lowered the

relative imaginary permittivity by 15 % from -100 °C to 200 °C. The system showed the presence of both α and β

relaxations corresponding to epoxy network. α transition corresponding to the PEI inclusion is not fully recognized
 ACCEPTED MANUSCRIPT
10

due to the limitation of the experimental temperature at 230 °C.

Loss values tested in all frequency and temperature ranges are represented in a 3D plot Fig. 8(b). The value of the

losses ε''r decreased slightly below Tα epoxy upon the addition of PEI. First relaxation whose maximum is around -

50°C is supposed to be the β relaxation of the epoxy network, where it is shifted with the variation of frequency.

PT
When the temperature approaches the glass transition temperature of the epoxy system a huge increase in the loss

values is identified. The beginning of this increase is due to α relaxation of the epoxy network, related to relaxation

RI
processes of the main chains of the cross-linked epoxy chains, where it is partially mask due to the increase of the

concentration of charge carriers. This relaxation arises from the fact that free charges are now immobilized in the

SC
material, and at sufficient high temperature, the charges can migrate in the presence of applied electric fields. As the

temperature increase above the glass transition temperature, the conductivity of the system increases giving rise the

U
increase of the concentration of charge carriers. This increase is seen in DM as well as DM10PEI. This increase is

relatively higher in the 10 wt.% PEI/epoxy blend where Maxwell-Wagner-Sillars (MWS) polarization can interfere
AN
as the process of (MWS) polarization exist in heterogeneous dielectrics, and is due to a delay in charge transfer at the

interface between components with different dielectric permittivity [13].

M

Regarding other studies of micro and nano filled epoxy networks, dielectric properties is studied for systems
D

filled with inorganic fillers, which is substantially different with the present system were the epoxy is blended with
TE

an organic PEI thermoplastic. It is noticed that the majority of the fillers have permittivity higher than that of the

epoxy networks. According to literature, when micro-fillers are added to the epoxy network, the permittivity of the
EP

final epoxy micro-composite is higher than that of the pure epoxy network [11,12,47,48]. It's noteworthy that the

incorporation of small percentages of nano-fillers (<2.5 vol.%) in the presence of micro-fillers decrease the value of
C

the epoxy micro-composite; however the real relative permittivity value as well as its imaginary relative permittivity
AC

still have higher values than the pure epoxy system [11,12].

The only way to decrease the permittivity value of the epoxy composite below the pure epoxy system was to add

nano-fillers with small critical values depending on their nature [11,15,16,47-49].These studies showed that the real

and imaginary parts of the relative permittivity of the epoxy nano-composite have increased under the influence of

nano-fillers above a small critical value, such as 0.5-1 vol.% of hexagonal boron nitride (hBN), TiO2, Al2O3 and ZnO

[14,49], 2.5 vol. % of barium titanate (BTO), calcium copper titanate (CCTO) and nano-silica [11,16], 5 vol. %
 ACCEPTED MANUSCRIPT
11

fumed silica and cubic boron nitride (cBN) [15,47]. Below these critical values the permittivity of the epoxy nano-

composite decreases along with different responses in the loss factor. For example, the losses found to be stable

using 20 nm nano-sized silica [11], higher for fumed silica, nano-sized BTO and CCTO, and 70 nm hBN [14-16] and

relatively lower using 20-70 nm TiO2, Al2O3, ZnO and cBN [47-49]. All the presented critical values were much

PT
smaller than the 9.8 vol. % of organic PEI blended with the epoxy network used in our study. In comparison with the

work of Fothergill et al. where they used 10 wt.% of TiO2 with an average size of 23 nm, the relative permittivity

RI
response was highly elevated due to the percolation phenomenon [50]. In the present study, the introduction of

micro-fillers into the epoxy network leads to a decrease of the relative permittivity ε'r without increasing losses,

SC
behavior in total opposition to previous works on inorganic fillers. The presence of thermoplastic in the system

proceeded in restricting the molecular dipolar movement of the epoxy system and thus decreases as well the relative

U
permittivity response of the epoxy blend. We observed the decrease with the addition of PEI under all variations of

temperature and frequency as seen in Fig. 8(a)

AN
3.5. Effective permittivity
M

Simulation of the effective permittivity of the DM10PEI dielectric response was calculated from the experimental

response of the pure systems of epoxy and PEI. The model, derived and studied by many authors [51-54], is an
D

analytical formula of the global dielectric response of composite material consisting of a host matrix filled with

ellipsoidal particles. The nodules are randomly oriented with respect to a given applied electric field as observed in
TE

the SEM image; Fig. 2(b) The effective permittivity of a medium having ellipsoidal inclusions can be expressed as

EP

∗ ∗
∗ ∗
= 1 + ∅ ∑" #,%,& ∗ ( ∗ ∗) ' (1)
!

∗ ∗ ∗
, and
C

where represent the calculated relative permittivity, experimental relative permittivity of PEI and the

experimental relative permittivity of the epoxy network respectively. The real and imaginary parts of the complex
AC

relative permittivity are separated using computer software. A schematic representation of a PEI nodule in the epoxy

network is shown in Fig. 9. The volume fraction ∅=((4/3)(πabc))/(lwh) of PEI in the epoxy network is, in our studied

case, equal to 9.8 vol.% where the densities of the DGEBA, MDEA and PEI are 1.15, 1.35 and 1.28 g/cm3

respectively. Lk is the depolarization factor in the direction characterized by index k (x, y, z). A standard result of

electrostatics gives [54]:

 ACCEPTED MANUSCRIPT
12

)*+ 0 -.
(# = , (2)
(. ) )/(. ) )(. * )(. + )

Where a, b and c denote the semiaxes of the ellipsoid. As discussed earlier the average diameter of the PEI

nodules into the epoxy network is 2-3 µm and thus the average major radius (a) is taken as 1-1.5 µm. The nodules are

PT
in an ellipse form having an average minor radius (b) = 4a/5 the major radius(a) and the second minor radius (c) is

considered to be in the same size as (b); taking into consideration that the size of the particles is measured in 2D

RI
plane. For the studied epoxy-PEI blend, L is numerically calculated using (a) of value between 1 and 1.5 leading to a

value of 0.275 and 0.362 for Lx and Ly respectively and Ly is equal to Lz.

SC
SEM images showed that ellipsoidal nodules were regularly dispersed into the epoxy network. Furthermore, the

U
system is considered to be randomly oriented with respect to the applied field. According to this fact, the summation
AN
of the three depolarization factors is applied in the proposed model. The simulated curve is calculated from the

experimental permittivity of DM and PEI measured at 1 kHz and is presented in Fig. 10. The simulated curve is

similar to the experimental one with a shift to higher values by an average of 9 %. This shift can be explained in
M

several points of view. From the theoretical work of Balberg et al., it can be suggested that excluded volume in PEI

can be at the origin of this discrepancy, as the dimensions from a 2D plane have been taken into account, [55]. In
D

addition to that and as blending of polymers might generate additional free volume as discussed before, it can affect
TE

the simulated value [45]. The increase in the free volume can decrease the density of the PEI nodules generating a

decrease in the relative permittivity value and thus can affect the simulated value. From the TEM image in Fig. 3, an
EP

interphase exists between the PEI nodule and the epoxy system. This interphase can be remaining low molar masses

epoxy-amine dissolved in the PEI phase and can also contain free volume as discussed previously. The presence of
C

epoxy-diamine dissolved in a PEI phase can form a system that have different permittivity value except that it cannot
AC

affect the relative permittivity value of the blend as this interphase have relatively small percentage in comparison

with the present PEI nodules and it is not present on all the nodules. Free volume that might be present in the PEI

nodules will make a system which have lower density and thus can change the effect of PEI nodules on the mixture's

the permittivity value.

Moreover, And finally, as the effective permittivity is not only affected by the fraction volume and the

permittivity of the different constituents, proper determination of the form of the particle–particle and particle–
 ACCEPTED MANUSCRIPT
13

network interaction potentials is necessary for further computing [56]. There are several factors that can affect the

relative permittivity value of an epoxy blend and thus additional studies can be further suggested in this domain.

4. CONCLUSION

PT
In this paper, the effect of phase separation phenomenon in the epoxy/thermoplastic blend on the thermal,

mechanical and electrical properties was examined. The phase separation phenomenon taking place between the

RI
epoxy network and PEI nodules was illustrated by SEM imaging and confirmed by having two different transition

processes in the DM10PEI system by DSC and DMA measurements. TEM images explored the presence of an

SC
interphase that surrounds PEI nodules; this cannot explain the presence of the shift taking place in the value of α

relaxation temperature corresponding to the PEI inclusions. This interphase might correspond to remaining low

molar masses of epoxy-amine dissolved in the PEI phase. However, the presence of free volume in the PEI nodules,

U
which decreases the density of the material, can change the aspect of the PEI influencing the permittivity of the
AN
mixture. The addition of PEI into the epoxy system did not affect either the storage modulus or the thermal

resistance under air or nitrogen atmosphere. Despite the fact, the real and imaginary parts of the relative permittivity
M

of the whole system decreased by 15 % in all the temperature and the frequency ranges. This decrease is also

investigated by means of simulation and it is seen that a shift of 9 % from the experimental values was observed
D

which can be due to several phenomena as the existence of excluded volumes that can't be visualized via 2D SEM
TE

image, and the presence of free volumes from the fabrication process as discussed.
EP

This novel proposed material with phase separation phenomenon is a good candidate for electrical insulation as

the real and imaginary relative permittivity values have decreased in all temperature values up to 230 °C.
C

ACKNOWLEDGEMENTS
AC

The authors wish to thank Pierre Alcouffe for realizing the TEM image at the Centre Technologique des

Microstructures CTµ of the University of Lyon and the CLYM (Centre LYonnais des Microscopies) for giving us

the access to use the ESEM. The French Ministry of Education and Research is gratefully recognized for providing a

grant for this project and the Ecole Doctorale Electronique Electrotechnique Automatique de Lyon ED160 for their
 ACCEPTED MANUSCRIPT
14

financial support.

REFERENCES
[1] Tagare DM. Risks, Operation, and Maintenance of Hydroelectric Generators. Electric Power Generation, John Wiley & Sons, Inc.; 2011, p.
15–43.
[2] Sichel EK, Gittleman JI, Sheng P. Electrical properties of carbon-polymer composites. JEM 1982;11:699–747. doi:10.1007/BF02672392.
[3] Barber P, Balasubramanian S, Anguchamy Y, Gong S, Wibowo A, Gao H, et al. Polymer Composite and Nanocomposite Dielectric
Materials for Pulse Power Energy Storage. Materials 2009;2:1697–733. doi:10.3390/ma2041697.

PT
[4] Phuchaduek W, Jamnongkan T, Rattanasak U, Boonsang S, Kaewpirom S. Improvement in physical and electrical properties of poly(vinyl
alcohol) hydrogel conductive polymer composites. J Appl Polym Sci 2015;132:n/a-n/a. doi:10.1002/app.42234.
[5] Van Velthem P, Ballout W, Horion J, Janssens Y-A, Destoop V, Pardoen T, et al. Morphology and fracture properties of toughened highly
crosslinked epoxy composites: A comparative study between high and low Tg tougheners. Composites Part B: Engineering 2016;101:14–20.
doi:10.1016/j.compositesb.2016.06.076.

RI
[6] Gupta AK, Harsha SP. Analysis of mechanical properties of carbon nanotube reinforced polymer composites using multi-scale finite element
modeling approach. Composites Part B: Engineering 2016;95:172–8. doi:10.1016/j.compositesb.2016.04.005.
[7] Zhao Y-H, Zhang Y-F, Bai S-L, Yuan X-W. Carbon fibre/graphene foam/polymer composites with enhanced mechanical and thermal
properties. Composites Part B: Engineering 2016;94:102–8. doi:10.1016/j.compositesb.2016.03.056.

SC
[8] Lin J-H, Lin Z-I, Pan Y-J, Huang C-L, Chen C-K, Lou C-W. Polymer composites made of multi-walled carbon nanotubes and graphene
nano-sheets: Effects of sandwich structures on their electromagnetic interference shielding effectiveness. Composites Part B: Engineering
2016;89:424–31. doi:10.1016/j.compositesb.2015.11.014.
[9] Tanaka T. Dielectric nanocomposites with insulating properties. IEEE Transactions on Dielectrics and Electrical Insulation 2005;12: 914–
28. doi:10.1109/TDEI.2005.1522186.
[10] Tsekmes IA, Morshuis PHF, Smit JJ, Kochetov R. Enhancing the thermal and electrical performance of epoxy microcomposites with the

U
addition of nanofillers. IEEE Electrical Insulation Magazine 2015;31:32–42. doi:10.1109/MEI.2015.7089120.
[11] Iyer G, Gorur RS, Richert R, Krivda A, Schmidt LE. Dielectric properties of epoxy based nanocomposites for high voltage insulation. IEEE
Transactions on Dielectrics and Electrical Insulation 2011;18:659–66. doi:10.1109/TDEI.2011.5931050.
AN
[12] Castellon J, Nguyen HN, Agnel S, Toureille A, Frechette M, Savoie S, et al. Electrical properties analysis of micro and nano composite
epoxy resin materials. IEEE Transactions on Dielectrics and Electrical Insulation 2011;18:651–8. doi:10.1109/TDEI.2011.5931049.
[13] Tsangaris GM, Kouloumbi N, Kyvelidis S. Interfacial relaxation phenomena in particulate composites of epoxy resin with copper or iron
particles. Materials Chemistry and Physics 1996;44:245–50. doi:10.1016/0254-0584(96)80063-0.
[14] Tsekmes IA, Kochetov R, Morshuis PHF, Smit JJ. The role of particle distribution in the dielectric response of epoxy–boron nitride
M

nanocomposites. J Mater Sci 2014;50:1175–86. doi:10.1007/s10853-014-8674-5.

[15] Katayama J, Ohki Y, Fuse N, Kozako M, Tanaka T. Effects of nanofiller materials on the dielectric properties of epoxy nanocomposites.
IEEE Transactions on Dielectrics and Electrical Insulation 2013;20:157–65. doi:10.1109/TDEI.2013.6451354.
[16] Tuncer E, Sauers I, James DR, Ellis AR, Paranthaman MP, Aytuğ T, et al. Electrical properties of epoxy resin based nano-composites.
Nanotechnology 2007;18:25703. doi:10.1088/0957-4484/18/2/025703.
D

[17] Han Z, Wood JW, Herman H, Zhang C, Stevens GC. Thermal Properties of Composites Filled with Different Fillers. Conference Record of
the 2008 IEEE International Symposium on Electrical Insulation, 2008. ISEI 2008, 2008, p. 497–501. doi:10.1109/ELINSL.2008.4570381.
[18] Heid T, Frechette M, David E. Nanostructured epoxy/POSS composites: High performance dielectrics with improved corona resistance and
TE

thermal conductivity, IEEE; 2014, p. 316–9. doi:10.1109/EIC.2014.6869400.

[19] Preetha P, Thomas MJ. AC breakdown characteristics of epoxy nanocomposites. IEEE Transactions on Dielectrics and Electrical Insulation
2011;18:1526–34. doi:10.1109/TDEI.2011.6032821.
[20] Maiez-Tribut S, Pascault JP, Soulé ER, Borrajo J, Williams RJJ. Nanostructured Epoxies Based on the Self-Assembly of Block
Copolymers:  A New Miscible Block That Can Be Tailored to Different Epoxy Formulations. Macromolecules 2007;40:1268–73.
doi:10.1021/ma062185l.
EP

[21] Ritzenthaler S, Court F, Girard-Reydet E, Leibler L, Pascault JP. ABC Triblock Copolymers/Epoxy−Diamine Blends. 2. Parameters
Controlling the Morphologies and Properties. Macromolecules 2003;36:118–26. doi:10.1021/ma0211075.
[22] Rebizant V, Venet A-S, Tournilhac F, Girard-Reydet E, Navarro C, Pascault J-P, et al. Chemistry and Mechanical Properties of Epoxy-
Based Thermosets Reinforced by Reactive and Nonreactive SBMX Block Copolymers. Macromolecules 2004;37:8017–27.
doi:10.1021/ma0490754.
C

[23] Girard-Reydet E, Vicard V, Pascault JP, Sautereau H. Polyetherimide-modified epoxy networks: Influence of cure conditions on
morphology and mechanical properties. J Appl Polym Sci 1997;65:2433–45. doi:10.1002/(SICI)1097-4628(19970919)65:12<2433::AID-
APP15>3.0.CO;2-1.
AC

[24] Williams RJJ, Rozenberg BA, Pascault J-P. Reaction-induced phase separation in modified thermosetting polymers. Polymer Analysis
Polymer Physics, Springer Berlin Heidelberg; 1997, p. 95–156.
[25] Pascault JP, Williams RJJ, Formulation and Characterization of Thermoset/Thermoplastic Blends , in Polymer Blends: Formulation and
Performance, D. R. Paul, C. B. Bucknall.; Wiley; 2000,p. 379-415. Chapter 13.
[26] Hedrick JL, Yilgör I, Wilkes GL, McGrath JE. Chemical modification of matrix Resin networks with engineering thermoplastics. Polymer
Bulletin 1985;13:201–8. doi:10.1007/BF00254652.
[27] Oyanguren PA, Riccardi CC, Williams RJJ, Mondragon I. Phase separation induced by a chain polymerization: Polysulfone-modified
epoxy/anhydride systems. J Polym Sci B Polym Phys 1998;36:1349–59. doi:10.1002/(SICI)1099-0488(199806)36:8<1349::AID-
POLB9>3.0.CO;2-M.
[28] Yang G, Zheng B, Yang J-P, Xu G-S, Fu S-Y. Preparation and cryogenic mechanical properties of epoxy resins modified by
poly(ethersulfone). J Polym Sci A Polym Chem 2008;46:612–24. doi:10.1002/pola.22409.
[29] Blanco I, Cicala G, Faro CL, Recca A. Development of a toughened DGEBS/DDS system toward improved thermal and mechanical
properties by the addition of a tetrafunctional epoxy resin and a novel thermoplastic. J Appl Polym Sci 2003;89:268–73.
doi:10.1002/app.12179.
 ACCEPTED MANUSCRIPT
15

[30] Hedrick JL, Yilgor I, Jurek M, Hedrick JC, Wilkes GL, McGrath JE. Chemical modification of matrix resin networks with engineering
thermoplastics: 1. Synthesis, morphology, physical behaviour and toughening mechanisms of poly(arylene ether sulphone) modified epoxy
networks. Polymer 1991;32:2020–32. doi:10.1016/0032-3861(91)90168-I.
[31] Bucknall CB, Gilbert AH. Toughening tetrafunctional epoxy resins using polyetherimide. Polymer 1989;30:213–7. doi:10.1016/0032-
3861(89)90107-9.
[32] Girard-Reydet E, Sautereau H, Pascault JP, Keates P, Navard P, Thollet G, et al. Reaction-induced phase separation mechanisms in modified
thermosets. Polymer 1998;39:2269–79. doi:10.1016/S0032-3861(97)00425-4.
[33] Cho JB, Hwang JW, Cho K, An JH, Park CE. Effects of morphology on toughening of tetrafunctional epoxy resins with poly(ether imide).
Polymer 1993;34:4832–6. doi:10.1016/0032-3861(93)90005-U.
[34] Hourston DJ, Lane JM. The toughening of epoxy resins with thermoplastics: 1. Trifunctional epoxy resin-polyetherimide blends. Polymer

PT
1992;33:1379–83. doi:10.1016/0032-3861(92)90110-I.
[35] Bonnet A, Pascault JP, Sautereau H, Taha M, Camberlin Y. Epoxy−Diamine Thermoset/Thermoplastic Blends. 1. Rates of Reactions before
and after Phase Separation. Macromolecules 1999;32:8517–23. doi:10.1021/ma981754p.
[36] Riccardi CC, Borrajo J, Williams RJJ, Girard-Reydet E, Sautereau H, Pascault JP. Thermodynamic analysis of the phase separation in
polyetherimide-modified epoxies. J Polym Sci B Polym Phys 1996;34:349–56. doi:10.1002/(SICI)1099-0488(19960130)34:2<349::AID-

RI
POLB16>3.0.CO;2-J.
[37] Gilbert AH, Bucknall CB. Epoxy resin toughened with thermoplastic. Makromolekulare Chemie Macromolecular Symposia 1991;45:289–
98. doi:10.1002/masy.19910450129.
[38] Bonnet A, Lestriez B, Pascault JP, Sautereau H. Intractable high-Tg thermoplastics processed with epoxy resin: Interfacial adhesion and
mechanical properties of the cured blends. J Polym Sci B Polym Phys 2001;39:363–73. doi:10.1002/1099-0488(20010201)39:3<363::AID-

SC
POLB1008>3.0.CO;2-Q.
[39] Girard-Reydet E, Riccardi CC, Sautereau H, Pascault JP. Epoxy-Aromatic Diamine Kinetics. Part 1. Modeling and Influence of the Diamine
Structure. Macromolecules 1995;28:7599–607. doi:10.1021/ma00127a003.
[40] Bucknall CB, Yoshii T. Relationship between structure and mechanical properties in rubber-toughened epoxy resins. Brit Poly J
1978;10:53–9. doi:10.1002/pi.4980100110.

U
[41] MacKinnon AJ, Jenkins SD, McGrail PT, Pethrick RA. A dielectric, mechanical, rheological and electron microscopy study of cure and
properties of a thermoplastic-modified epoxy resin. Macromolecules 1992;25:3492–9. doi:10.1021/ma00039a029.
[42] Levchik SV, Weil ED. Thermal decomposition, combustion and flame-retardancy of epoxy resins—a review of the recent literature. Polym
AN
Int 2004;53:1901–29. doi:10.1002/pi.1473.
[43] Heux L, Halary JL, Lauprêtre F, Monnerie L. Dynamic mechanical and 13C n.m.r. investigations of molecular motions involved in the β
relaxation of epoxy networks based on DGEBA and aliphatic amines. Polymer 1997;38:1767–78. doi:10.1016/S0032-3861(96)00694-5.
[44] Wan J, Li C, Bu Z-Y, Xu C-J, Li B-G, Fan H. A comparative study of epoxy resin cured with a linear diamine and a branched polyamine.
Chemical Engineering Journal 2012;188:160–72. doi:10.1016/j.cej.2012.01.134.
[45] Akay M, Cracknell JG. Epoxy resin–polyethersulphone blends. J Appl Polym Sci 1994;52:663–88. doi:10.1002/app.1994.070520509.
M

[46] Choudhury A. Dielectric and piezoelectric properties of polyetherimide/BaTiO3 nanocomposites. Materials Chemistry and Physics
2010;121:280–5. doi:10.1016/j.matchemphys.2010.01.035.
[47] Heid T, Fréchette M, David E. Epoxy/BN micro- and submicro-composites: dielectric and thermal properties of enhanced materials for high
voltage insulation systems. IEEE Transactions on Dielectrics and Electrical Insulation 2015;22:1176–85. doi:10.1109/TDEI.2015.7076820.
D

[48] Nelson JK, Fothergill JC. Internal charge behaviour of nanocomposites. Nanotechnology 2004;15:586. doi:10.1088/0957-4484/15/5/032.
[49] Singha S, Thomas MJ. Dielectric properties of epoxy nanocomposites. IEEE Transactions on Dielectrics and Electrical Insulation
2008;15:12–23. doi:10.1109/T-DEI.2008.4446732.
[50] Fothergill JC, Nelson JK, Fu M. Dielectric properties of epoxy nanocomposites containing TiO2, Al2O3 and ZnO fillers. 2004 Annual
TE

Report Conference on Electrical Insulation and Dielectric Phenomena, 2004. CEIDP ’04, 2004, p. 406–9.
doi:10.1109/CEIDP.2004.1364273.
[51] Polder D, van Santeen JH. The effective permeability of mixtures of solids. Physica 1946;12:257–71. doi:10.1016/S0031-8914(46)80066-1.
[52] Fricke H. The Maxwell-Wagner Dispersion in a Suspension of Ellipsoids. J Phys Chem 1953;57:934–7. doi:10.1021/j150510a018.
[53] Altshuller AP. The Shapes of Particles from Dielectric Constant Studies of Suspensions. J Phys Chem 1954;58:544–7.
EP

doi:10.1021/j150517a008.
[54] Asami K. Characterization of heterogeneous systems by dielectric spectroscopy. Progress in Polymer Science 2002;27:1617–59.
doi:10.1016/S0079-6700(02)00015-1.
[55] Balberg I, Anderson CH, Alexander S, Wagner N. Excluded volume and its relation to the onset of percolation. Phys Rev B 1984;30:3933–
43. doi:10.1103/PhysRevB.30.3933.
C

[56] Brosseau C. Modelling and simulation of dielectric heterostructures: a physical survey from an historical perspective. J Phys D: Appl Phys
2006;39:1277. doi:10.1088/0022-3727/39/7/S02.
AC
 ACCEPTED MANUSCRIPT
16

Tables:

PT
Table I: Chemical Structures of the used reactants

12

RI
Reactants Formula Supplier
-1
(g.mol )

SC
DGEBA Huntsman 382

U
MDEA LONZA 310
AN
PEI GE 26,000
M
D

Table II: Temperatures (Tα1, Tα2), amplitudes (h1, h2) obtained from Tan δ curves Fig. 5(b), and storage modulus for DM, DM10PEI and PEI.
TE

Tα (°C) Amplitude (h)

h2/h1
Tα1 Tα2 h1 h2
EP

DM 183 0.62 -

DM10PEI 183 220 0.57 0.08 0.14

C

PEI 226 2.58 -

AC

Figures:

Fig. 1: Scheme presenting different morphological type of phase separation; left to right: particulate, bi-continuous
and phase inverted.
Fig. 2: SEM image of DM (a) DM10PEI blend (b) and Size distribution of the PEI nodules into the epoxy network (c)

Fig. 3: TEM image of DM10PEI

 ACCEPTED MANUSCRIPT
17

Fig. 4: Weight loss as a function of temperature for DM, DM10PEI and PEI a) in nitrogen atmosphere; b) in air atmosphere.

Fig. 5: DSC thermograms of DM, DM10PEI and PEI and b) Derivative of heat flow with respect to temperature of DM10PEI

Fig. 6: Storage Modulus E’ (a) and loss factor Tan δ (b) as function of temperature of DM, DM10PEI and PEI at 1 Hz.

Fig. 7:Relative permittivity, b) Dielectric loss and zoom on the dielectric loss; as function of temperature at 1 kHz for DM, DM10PEI and PEI

Fig. 8: Relative permittivity and b) Dielectric loss; of DM and DM10PEI versus temperature and frequency

PT
Fig. 9: scheme of the PEI nodule into the epoxy network

Fig. 10: experimental permittivity at 1 kHz for DM, DM10PEI and PEI; effective simulated permittivity of DM10PEI

RI
U SC
AN
M
D
TE
C EP
AC
 ACCEPTED MANUSCRIPT

PT
RI
U SC
AN
M
D
TE
EP
C
AC
 ACCEPTED MANUSCRIPT

PT
RI
U SC
AN
M
D
TE
EP
C
AC
 ACCEPTED MANUSCRIPT

PT
RI
U SC
AN
M
D
TE
EP
C
AC
 ACCEPTED MANUSCRIPT

PT
RI
U SC
60
AN
M

50
Size distribution

D
TE

40
EP

30
C
AC

20

10

[1-2] [2-3] [3-4] [4-5]

Diameter of nodules (µm)

 ACCEPTED MANUSCRIPT

PT
RI
U SC
AN
M
D
TE
EP
C
AC
 ACCEPTED MANUSCRIPT

PT
RI
U SC
AN
M
D
TE
EP
C
AC
 ACCEPTED MANUSCRIPT

PT
RI
U SC
AN
M
D
TE
EP
C
AC
 ACCEPTED MANUSCRIPT

PT
RI
U SC
AN
M
D
TE
EP
C
AC
 ACCEPTED MANUSCRIPT

PT
RI
U SC
AN
M
D
TE
EP
C
AC
 ACCEPTED MANUSCRIPT

PT
RI
U SC
AN
M
D
TE
EP
C
AC
 ACCEPTED MANUSCRIPT

PT
RI
U SC
AN
M
D
TE
EP
C
AC
 ACCEPTED MANUSCRIPT

PT
RI
U SC
AN
M
D
TE
EP
C
AC
 ACCEPTED MANUSCRIPT

PT
RI
U SC
AN
M
D
TE
EP
C
AC
 ACCEPTED MANUSCRIPT

PT
RI
U SC
AN
M
D
TE
EP
C
AC
 ACCEPTED MANUSCRIPT

PT
RI
U SC
AN
M
D
TE
EP
C
AC
 ACCEPTED MANUSCRIPT

PT
RI
U SC
AN
M
D
TE
EP
C
AC
 ACCEPTED MANUSCRIPT

PT
RI
U SC
AN
M
D
TE
EP
C
AC